@attoparsec He's not the only one: https://spacenews.com/china-files-itu-paperwork-for-megaconstellations-totaling-nearly-200000-satellites/

@AkaSci Just in case anyone needs to see it: this can't work with today's or even near-future technology.

https://pivot-to-ai.com/2025/12/01/ai-data-centres-in-space-why-dcs-in-space-cant-work

#OrbitalAI #DataCentersInSpaceCantWork

@evacide There are a lot of such blogs; what would make yours distinctive and more worthwhile to what audience?

@pikesley see recursive pollution and model collapse. Eg https://berryvilleiml.com/2026/01/10/recursive-pollution-and-model-collapse-are-not-the-same/

But also see this paper arguing that collapse does not happen if you accumulate training data rather than replace old data with new data; instead, it converges towards a limited level of errors.
https://arxiv.org/pdf/2404.01413